Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: Tenchusatsu who wrote (181303)	5/21/2005 11:21:48 PM
From: Elmer Phud	Read Replies (1) \| Respond to of 186894

Ten In the interests of fairness and accuracy, the points Ali raises about process variation are real, but that was never the point. This discussion started when I quoted Intel as saying their 90nm yields were well into the "world class" range. Ali took exception with that statement, claiming that Intel's yields do not reflect the effects of these intra-die parametric variations. Implied in this claim by Ali is that Intel's numbers include an unknown number of defective parts which he believes were not screened by Intel's tests and therefore Intel's yield claims are misleading. The potential defect modes he described are real however they are very much subject to Intel's test screens. The objections he raised and the examples he used were clear indication of a outdated understanding of modern test capability. So to summarize, the yield challenges Ali raised regarding 90nm processing are real, however his understanding of Intel's (or anyone else's) ability to screen for these failure modes is sorely lacking. The claims Intel made of having "world class yields" are not flawed due to unscreened potential failure mechanisms inherent in 90nm processes as he claimed, because those defect modes are screened for and the yields reflect that.

To: Tenchusatsu who wrote (181303)	5/21/2005 11:38:21 PM
From: Dan3	Read Replies (1) \| Respond to of 186894

Re: no one would think of buying anything less than 3.8 GHz ... There was a time, a couple of years ago, when AMD's process/desgin had high absolute yields but low bin splits. It resulted in AMD's most expensive Athlon selling for around $100, and most of their chips selling for half that or less. At the same time, Intel's Pentiums were selling for 5 times that and more. jc-news.com; This was a problem for AMD, though you can't seem to understand why high yields with low bin splits can be an issue. A trend towards the reverse of that situation has now appeared, and if it continues for long enough, it will become a real issue for Intel.

To: Tenchusatsu who wrote (181303)	5/22/2005 3:44:58 PM
From: Ali Chen	Read Replies (1) \| Respond to of 186894

Tench, "no one would think of buying anything less than 3.8 GHz ..." You misunderstand, or intentionally shift the accent. That is not what I said (or meant). To my personal taste, no one would think of buying anything MORE than 2 - 2.4GHz, there is no practical urge. However, Intel needs to prevent an erosion of it's image as undisputed performance leader, and for this reason you need 4-4.4GHz chips NOW, even if you can supply them only to reviewers or for SPEC submissions. It looks like you can't make them currently, period. "the "proof" of your assertions come from your knowledge of "transistor variability" ..." It was not the point, I was trying to help Elmer to understand a simple thingy that being at "defect density" below certain point is not a guarantee of desired spectrum of otherwise functional products. That's why I like to bring back the classic example of "One Million CuMines Per Week". With two generations down the technology road, new effects give more troubles for sort at wafer level, not less. "Possibility doesn't equal proof." Who gives a duck. I just tried to help you to set proper expectations about real meaning of "world-class defect density", especially for future "65nm node". Cheers, - Ali